A general multimedia representation space model toward event-based collective knowledge management

Abstract

Emergent technologies such as smart phones and wireless Internet have transformed the Web from a static data publishing platform into a collaborative information sharing environment. Yet, attaining the next stage in Web engineering, i.e., the so-called Intelligent Web: allowing meaningful human-machine and machine-machine collaboration, requires another breakthrough: allowing the sharing and organization of collective knowledge (CK), where CK underlines the combination of all known data, information, and meta-data concerning a given concept or event. In this context, various methods have been put forward to perform automatic event extraction and description. Yet, most of them do not capture the semantic meaning embedded in Web-based multimedia data, which are usually highly heterogeneous and unstructured. To address this problem, we introduce in this study a generic Multimedia Representation Space Model called MRSM, designed for multimedia data and multimedia-based event representation, in order to allow event detection and identification based on multimedia CK. We formally define MRSM, its dimensions, their coordinates, and the associated distance (similarity) metrics and properties. We then provide the building blocks for an Eventbased Collective Knowledge (CK) Management Framework, built upon MRSM, and geared toward effective CK management. The proposed approach provides a means of extracting, representing, and linking events from heterogeneous multimedia data without any prior knowledge about event-related clues. Preliminary tests confirm the quality and potential of our approach.